Twist's Review #10 - New Passivation Structure for GaN HEMT

Paper: Novel Stacked Passivation Structure for AlGaN/GaN HEMTs on Silicon With High Johnson’s Figures of Merit

As I have already mentioned before, GaN is quite a popular material these days along with SiC in the field of power electronics. However, their implementation does not stop there, but also includes RF devices. Now I am not actually too familiar with RF, but regardless I think it is important to dive into it as to not ignore GaN’s potential for other applications. Better than never right?

This paper was written by a group from South China University of Technology, and showed a new way to do passivation in AlGaN/GaN HEMTs. As far as I know passivation has been done “traditionally”, where an insulating layer is deposited around the terminals. So far I have seen SiO2, HfO2, Al2O3, SiN, and AlN being used as the passivation layer.

Here, the authors propose the usage of two passivation materials in tandem, where HfO2 is deposited first, and later SiO2 in a stacked passivation structure (Sample C). They compared the performance of this structure with devices passivated with only HfO2 (Sample A) and only SiO2 (Sample B). Sample A showed a relatively high breakdown voltage of 112 V, B of 20 V and C of 140V. This new stacked passivation structure helped with this increase in breakdown voltage as the transition between the HfO2 and SiO2 shows a parabola-like curve of electric field, where it drops towards the end of the HfO2 and peaks again at the beginning of SiO2. This makes for two peaks, one at the beginning of HfO2 and another at the SiO2. This other peak is said to be able to modulate nearby electric field to reduce the electric field at the drain edge of the gate.

Unfortunately sample C has an inferior RF performance compared to sample B due to the higher parasitic capacitance, a consequence of high dielectric constant. This though, does not mean that sample C is not an achievement. The Johnson’s Figure of Merit (J-FOM) is a metric used to measure the performance of RF devices, formulated as the product of the breakdown voltage and current gain cutoff frequency of the device (higher is better). The J-FOM of sample C turns to be higher compared to both sample A and B coming out to 4.4 THzV, owing to its more well-roundedness.

Reference:
[1] X. Liu, J. Qin, J. Chen, J. Chen and H. Wang, “Novel Stacked Passivation Structure for AlGaN/GaN HEMTs on Silicon With High Johnson’s Figures of Merit,” in IEEE Journal of the Electron Devices Society, vol. 11, pp. 130-134, 2023, doi: 10.1109/JEDS.2023.3241306.
keywords: {Passivation;Logic gates;Electric fields;MODFETs;HEMTs;Hafnium oxide;Radio frequency;AlGaN/GaN HEMTs;breakdown voltage;stacked passivation layer},

I saw this paper as I was browsing some journals in IEEE and came across the Journal of the Electron Devices Society. The journal was fully open access, which was already exciting, but it also seemed to contain quite insightful research. Upcoming reviews would likely come from this journal as I found some other interesting papers. I do though apologize if the topic has been a bit more monotonic because this is my background, so I do get excited when I see new research around the topics I have learnt before but I will try my best to get around other topics.

Anyway, I thought this paper was interesting as they introduced truly a novel passivating structure for GaN HEMTS (or at least as far as I have seen). The stacking of different dielectric materials is quite an interesting concept and it seems like there are other avenues for exploration here such as using SiN as one of the materials as it had seen quite a success in previous research. Other factors such as the length of the HfO2 within the stacked layer may also be interesting to explore.